is link(2) is not included in the “system calls” section of man.freebsd.org/cgi/man.cgi?query=symlink&sektion=7

mjacob: 1) the symlink itself 2) symlink(7) focuses on symbolic links, while link(2) focuses on hard links, i agree that's confusing

where's some meaningful documentation for the read(1) builtin command?

I'm trying to eliminate a dependency on bash

probably in each of the manpages for the various shells

dgriffi: sh(1) has a decent description, don't try to look at read(1)

imo builtin(1) is a load of bullshit but I don't think I'd get any buy-in for what I'd prefer

I'm trying to get the functionality of bash's "read" builtin with its -d option.

bash(1) has a description of its 'read' with -d

using bash is not an option here, I suppose?

all good citizens de-bash what they can :-p

builtin(1) is still a load of bullshit, even if what they really wanted is bash' description

I've been told to not admit too loud that I'm a BASH fetishist

kevans: I'm trying to eliminate the use of "read -d"

I do like using bash interactively, but for what I'm working on, I don't want to have to tell people "you must install bash"

it's looking like I'll have to port this script to perl

Well, now you're telling people "you must install perl". :3

oh.. right

I'm still WAY too deep in love with BASH...

It's not my interactive shell but if I need to create a script to be shared with others, it's my go to.

My interactive shell is ZSH.

It just happened to work a tiny bit better than BASH in my case, less surprises, syntax highlighting, well there was a bit of benefits, yea

I ended up with some 200-300 lines ZSH configuration file that does everything that I could ever ask for

Like yea without any framework whatsoever

That's a lot of reconstruction, re-structuring, refactoring (I guess that's the term) but it's so worth it

i don't mind bash as much if the shebang is properly written

my #1 annoyance with it is that it doesn't really put any effort into disabling some of the bash-isms if executed as 'sh'

screw it. I'll just like bash as a requirement

list

I have some warnings about a corrupted GPT table on one of my zfs disk, does this happen when you dedicate a disk to ZFS without partitioning it and using a partition? I've been a bit hesitant to fix it for fear that doing so might overnight some ZFS data

crb: yes, but it's fine to have ZFS on unpartitioned drive, unless you want to boot from it

I mean it does seem to work fine and has for a long time, I guess I'm just thinking it might be safer to partition and then use the partition

I dislike having the warning come up, and for some reason I thought that ZFS put a fake partition table at the front of the disk even when you dedicate the whole disk

crb: Most applications would like/prefer a partition as a safeguard to make sure there isn't anything overwritten. However, they're certainly not required by any means. You can dedicate an entire disk to any FS if you want to.

No partitions needed. Unless, as mzar mentioned, you're trying to boot from it.

What, exactly, is producing the warning message you're seeing? Is this at boot time or something?

What type of disk is it (gpt? mbr?) and how is it being mounted?

If you can, just toss the data on that disk to somewhere else and then reformat again creating a partition (slice.) I'm not sure if that'll actually make any difference, but you might not see those errors anymore?

I've never experienced this, so I'm just guessing.

ek: GEOM produces them at boot

it's an SSD that I didn't partition but merely used in a zfs create like zfs create foo nda0

ek: as it's terabytes of info it's not really easy to just toss somewhere

Does FreeBSD have the ability to update its kernel "on the fly", meaning no reboot needed? Linux cannot, but I can't for the life of me remember doing this while using FreeBSD. Maybe I was always running an out of date kernel without realizing it? (Since full disk encryption was used, it didn't come back on its own if I just rebooted it remotely.)

Unit640, no

Hmm.

Unit640: Definitely not. But, kernel modules can rebuilt and re-loaded on-the-fly (usually.)

Unit640: Actually ksplice.oracle.com tuxcare.com/enterprise-live-patchin…elcare-enterprise-to-oracle-ksplice

Unit640: Ubuntu and RHEL/Clones can do live kernel patching

crb: I wouldn't worry about the boot message. It's likely just a safeguard or fail safe to alert someone about a possible issue with partition issues. Again, I've never seen this, but I can't fathom why that would be an actual issue.

I don't partition my data drives at all. Only my OS drive

Same. Never had any problems.

Although, I also don't generally use single ZFS drives. It's been a long time since I've done that.

Kinda has me curious now, though.

Oh, my! I didn't realize how late it was. I'd better get to sleep. Lots of prep tomorrow for the US holiday that follows.

armin: i already suspected that the symlink itself is hardlinked. but is there a referencable, quotable source for that?

armin: i still don’t understand why link(2) is not listed in symlink(7) as “does not follow symlinks”

is there an easy way to zfs send a whole set up filesystems under a parent?

never mind, thank you

SponiX: Well, I use Debian Linux. Didn't think such major things differed between distros.

It bothers me that a server has to be manually "babysat", regardless of the OS running.

can i run 3rd os other than linux in jail?

luke_jobless_sb, I didn't think you could run linux in a jail, you can run linux in a VM like bhyve

luke_jobless_sb, as I understand it jails are not a VM

luke_jobless_sb: jails use the FreeBSD kernel and provide a restricted FreeBSD user space on the host machine

crb: not that i have used it, but docs.freebsd.org/en/books/handbook/jails/#creating-linux-jail

A linux userspace wouldn't be able to make use of the FreeBSD kernel (except maybe Gentoo but that's some dark magic I don't study)

luke_jobless_sb: If you're familiar with Linux's lxc project, think of it like that, lxc can act as a container for multiple distributions because they all use the Linux kernel but lxc can't make Windows containers.

i dont any meaningful linux server. but it seems it's time to try such container

Unit640: What do you mean by babysat, exactly?

There's plenty of logging/monitoring utilities and 99% of the time I just use ansible to handle something.

i ssh

lots of people do

luke_jobless_sb, ok but I think the difference there is that it's not a linux kernel running it's linux binaries running under the freebsd kernel

FreeBSD has special sys call support for running linux binaries. There aren't other OS that can be emulated in this way whereas a bhyve VM can run virtually any OS and that OS kernel is actually running

crb: I don't have any opinion.

crb: can i run haiku or darwin in freebsd containerization?

luke_jobless_sb: no.

luke_jobless_sb: you seem to misunderstand the concept of "containers".

Do those projects use FreeBSD kernels (I don't know if they actually do)?

They don't.

ah, then no

(... well, Darwin in some distant past was sorta kinda BSD related)

Alver: I think I misunderstand and this beer helped

You *might* be able to run Gentoo/FreeBSD (the Gentoo project but using the FreeBSD kernel, although I don't know the status of that project).

luke_jobless_sb: you can either run current or older FreeBSD versions in a jail, which can sorta kinda run Linux binaries if you want, or you can install a full OS in bhyve VMs, and then your CPU architecture is the limit.

Debian used to have a kFreeBSD branch. Doubt it gets much testing though.

But, even there, not in a jail.

that's a dayum shame

Eh, it's a "look mon I'm using a nonstandard kernel" kind of thing.

s/mon/mom/

lol

They have (had?) the same with GNU HURD.

Alver: so if there exists a linux project that runs all of these 3rd os binaries(which belongs to linux) in this manner, can i still run their binaries but it is still on bsd kernel?

"The development of Debian GNU/kFreeBSD has officially terminated as of July 2023 due to the lack of interest and volunteers"

very sad

luke_jobless_sb: a jail is using the same kernel as the host, period

You cannot "install" an OS in a jail. You can have a different userland, provided it can use the host kernel

So, short answer, no

If you want a different OS, use VMs. That's what they exist for. Jails have a different purpose

afaik FreeBSD can only run FreeBSD jails (for now, again I'm not sure of the status of the Gentoo/kFreeBSD project)

That being said, I was trying to get podman to work in a jail. Would have enabled "standard" docker images to run, which tends to be Linux oriented. Quite unsuccesfully

Alver: i just use jails for all. I think those you call purpose is at some point diminished. I just like the fact that I can try them on a new jail and delete it.

o_0

i also like that I can cut internet connection within the jail sometimes

luke_jobless_sb: are you telling us, when you want to use firefox or something you install it in a jail and use that?

You cannot use jails "for all".

TommyC: heh, that would be a fun thing to do, in a way.

Alver: i was thinking about my jails and they don't really fit in those purpose. they are just jailed. nothing so much sake of purposeful

*fullness

luke_jobless_sb: What are you, a docker dev?

TommyC: it's like api jail, nginx jail, client jail, ...

TommyC: you get the idea

git server jail, znc jail

yeah those makes sense

that's for example my uses of jails. pretty much what docker folks do

i feel lazy to 'exactly' try other OS

It's much more efficient doing it in jails too. VMs have overhead, both in resources and time spent in admin

Alver: yes. i am cheap.

luke_jobless_sb: you should call it "cost-efficiënt". :D

TommyC: By "babysit", I mean any kind of action that a human being has to do related to the server, as opposed to it applying all updates on its own in a fully automated manner.

I repeatedly get the feeling that many people don't want such a scenario/computer world because of job security/sense of purpose reasons.

Do... do you guys not automate that? :3

Automate the unautomateable?

Why are updates not automatable?

I feel like that's one of the first things that people automate.

Some servers are retarded. BIOS set to always on and one hp prolient dl360 for some reason ignores it

but that's hopefully an exception to normal

it also prompts for whether we want auto rebuild or not

on the something or other smart array 420 or so...

I think they're professional comedians

TommyC: Because kernel updates to Linux (and apparently FreeBSD) cannot be applied on the spot, while running, so a reboot is needed, which means downtime and getting stuck at the "enter decryption passphrase" thing. For one.

TommyC: And even with "automatic upgrades" enabled on Debian Linux, it doesn't actually auto-restart the services affected by the updates (except in some cases), so you still need a human to do that.

Unit640: I thought I linked you to the methods that provide that for Linux lol

SponiX: Still researching that, but the downtime for the reboot still would happen.

SponiX: The hacking of the boot process is scary.

Unit640: there is a commonly used service in linux debian based distributions called "unattended-upgrades" that has removed much of that manual intervention...

Well kernel live patching is available for Linux, It only really supports Ubuntu and RHEL. but it does exist

SponiX: kexec supports realisticlly any env properly setup for it..

Tenkawa: Yeah, but every time I've looked at it - It was too complicated for me. Oracle Ksplice was easy to use, and the Tuxcare support likely isn't that tough either

kexec would be cool to have here

SponiX: yeah I was going to mention it has a high learning curve

I just Unix as a hobby, so the once a week or so reboot doesn't bother my users too much

Ah.. I've been using Unix since 79

heheh

Tenkawa: then you are even OLDER than I am... I started playing with Slackware in about 1995

SponiX: yeah started on Slack in 93

Was running an ISP with it...

looks like last activity was in 2016

based on the github

kexec in linux sat for years too... just recently got popular again

his talk goes over a lot of the same stuff you would be expected to solve if you wrote your own bootloader

GDT, IDT, Page tables, TSS

so can I run other os or other os binaries?

other than linux?

i was just asked back and answered some questions(drunk), gone sleep. and it seems I just need to work it out myself

Tenkawa: I'm using "unattended-upgrades", like I mentioned. It doesn't automate it.

It very much requires "attention" from a human, in spite of the name.

Unit640: not from what I've seen..

ok, linux updates can be discussed in a different channel

Diablo-D3: Governments aren't strictly anti- but are acting on the extreme of "safe side". Everyone is trying to find a "no-trip approach", but several medicinal startups already bakrupted on this - they were trying to synthesize similar molecules but with no trip effect, or to find some trip stoppers (apart from known ethanol and bezno, which are stopping the trip immediately, which also results

in no meaningful effects from the unfinished trip), but it's all crap. The psio+SSRI=notrip will have no effects as well. I believe that it's the absolute chaos in serotonin receptors that "makes" the trip (pleasant or unpleasant experience - doesn't matter) which actually triggers the change in the following weeks. Numbing how the psilocybin affects the 5-HT1A receptor with SSRIs simply will make

it not work at all.

Apologies, wrong channel.

o_O

Wrong chan, as I said. That's a conversation related to Polish government's study on psilocybin on despression treatment, which they are planning to do with psilo along with SSRI. No drug abuse stuff here. My comment was supposed to go on #nutrition. I again apologise for being a dumbass after changing IRC clients ;-)

ssri stop psilo from working however

Yup, that's the topic discussed. Polish government is going however planning to spend some $4M to "test" this theory and they proudly call it "one of the world-first attempts at assessing the efficacy and tolerance of SSRI with psilocybin. It's a complete waste of public money. Apologies for Polish link but it's not available in English. Translate on your own:

cool, anyway, when do we get a working libuuid implementation?

does it do more than generate uuid's?

yeah, i mean, the current e2fsprogs implementation is missing a slew of features... bugs.freebsd.org/bugzilla/show_bug.cgi?id=251489

SponiX, Tenkawa: Out of curiosity, why did you two stop using Slackware?

TommyC: moved to Debian and HP-UX (for my job)

ah

TommyC: Think mine is because Linux Mint was more friendly to new users in the friends and family category. So, I started running it too in order to help them

I was a Sun/HP/Linux admin back then